Commissioning Levels and Integrated Systems Testing: Why the Schedule Usually Gets Them Wrong

Ask anyone who has built a data center what a commissioning agent does during design, and you'll get a reasonable answer. They read the drawings, they check the sequences of operations, they make sure the building can actually be tested once it's built. Design review is an accepted part of the CxA scope, and most owners require it.

Ask the same person whether the CxA reviewed the construction schedule before it was baselined, and you'll usually get a pause.

That pause is the subject of this article. A design can be perfectly commissionable and still land on a schedule that makes it impossible to commission well — and the clearest place that shows up is in how commissioning levels and integrated systems testing get represented on the schedule. The scheduler who built that schedule isn't wrong, they just don't have the information to see the problem. The CxA does, and on a growing number of projects, they also have something more useful than opinion: data from the last time.

Two different questions

Design review answers one question: can this building be commissioned? Are the test points there, is the controls architecture sensible, can the redundant systems be isolated and exercised without taking the facility down?

Schedule review answers a different one: can this building be commissioned in the time you've allowed, in the order you've laid out, with the resources you've assumed?

Those are separate questions, and a yes to the first says nothing about the second. A well-designed electrical system with a clean topology can still be scheduled so that every lineup energizes in the same week, with one Cx crew and one vendor technician expected to cover all of them. The drawings are fine. The schedule feels like a guess.

What commissioning levels actually are

Commissioning on large facilities generally moves through a set of levels. Definitions vary by owner and by Cx firm, and that variance is its own problem, but the shape is consistent:

Level 1 and 2 cover factory testing and delivery, receipt, and installation checks. Did the equipment arrive undamaged and get installed correctly?
Level 3 is startup and energization. Power goes on, the equipment runs for the first time, usually with the manufacturer's technician present.
Level 4 is functional performance testing. Each system is put through its sequences and proven to do what the design says.
Level 5 is integrated systems testing, or IST. Everything runs together, and the facility is pushed through failure scenarios to confirm it behaves the way the owner will depend on.

Each level depends on the one before it. On a data center, energization order is dictated by the electrical topology: utility, then medium voltage switchgear, then transformers, then downstream distribution, then the loads, with cooling running before anything that generates heat comes online. That chain doesn't care what the schedule says.

What a CxA sees in a schedule that a scheduler doesn't

A construction schedule is built around trades and installation. Concrete, steel, envelope, mechanical rough-in, electrical rough-in, and so on. Commissioning tends to get added near the end as a block of activities. Sometimes it seems like an afterthought.

A CxA looking at the same schedule tends to find some version of the following:

Energization out of order. Downstream gear scheduled before its upstream source is ready, or every lineup stacked into the same window with no allowance for the crew and vendor support each one actually needs.

No float against the commercial operations date. Startup and functional testing are scheduled as fixed-duration activities, back to back, ending exactly at the date the owner needs the building. The first slip anywhere upstream lands directly on integrated testing.

IST shown as a single activity. On the schedule it's one bar. In practice it's a multi-day event, sometimes multi-week, that requires the whole building quiet, every trade off the floor, and a room full of people watching screens. Scheduling it as one line item hides all of that.

Vendor startup missing or untied. The manufacturer's technician shows up as a note in the mechanical contractor's activities, if at all. The Cx work that depends on that startup has no logic tie to it.

Controls treated as a punch item. The building management system and its integration with every piece of equipment is the thing functional testing actually exercises. If it's scheduled as finishing work rather than as a predecessor to Level 4, testing will start before there's anything to test.

Milestones that don't map to the Cx workplan. The schedule says "Commissioning Complete, Building A." The Cx plan has forty checklists and six functional tests behind that phrase. When the schedule gets updated, the percent complete on that milestone is someone's guess.

None of this is malice. The scheduler is working with what they can see. But the Cx sequence deserves the same rigor as the rest of the schedule, and skipping it doesn't make the argument go away, it just moves it to later, when it's more expensive and everyone's expectations have already hardened. Which raises the question: why are CxAs still finding these things in the third schedule update instead of in the baseline?

Often, there was no CxA when the baseline was built

The answer is usually not that anyone forgot. It's that the timing was never going to work.

On a typical project, the general contractor is awarded and the baseline schedule is developed and approved as a contract deliverable in the first month or two. The commissioning agent is procured on a separate track, often by the owner, and that RFP frequently doesn't go out until the construction documents are complete. By the time the CxA is under contract, has been onboarded, and has read the drawings, the baseline is locked. The commissioning milestones in it are whatever the scheduler assumed, and the CxA's first job is to figure out how to live inside them.

This is the same problem as late design review, just one step downstream. A CxA who joins at 90% drawings can comment, but the decisions that mattered were made at 30%. A CxA who joins after the baseline is approved can comment, but the sequence and the durations are already contractual. The comments become change requests, and change requests lose.

So "put schedule review in the CxA scope" is necessary and not sufficient. If the contract is signed after the baseline is approved, the scope covers reviewing a document that can no longer be shaped. The fix is earlier procurement, or at minimum, someone with commissioning experience on the owner's side reviewing the baseline before it's approved, as a stand-in until the CxA is on board. Plenty of owner's reps and construction managers have that background. Very few are asked to use it for this.

Experience is an opinion. Data is evidence.

Here's where it gets interesting, and where the industry is changing.

A seasoned commissioning agent walks into a schedule review knowing, from a career's worth of projects, that functional testing always takes longer than the GC thinks. That's real knowledge and it's worth having in the room. But in the room, it's an opinion. And an opinion tends to lose to a scheduler with a Gantt chart and a contractual end date.

Now imagine a different version of the same conversation. The CxA says: on our last eleven hyperscale projects, mechanical functional testing ran about 1.6 times the baselined duration, and on nine of them the extra time came from controls integration issues that weren't visible until the sequences were exercised. On seven of the eleven, the first energization slipped by an average of twelve working days, and in every one of those cases the cause was upstream of the gear itself.

Those figures are illustrative, but the shape of them is not. Cx firms with real records from past projects can put numbers against exactly the things schedules get wrong:

Actual duration of startup and functional testing by system type, against what was planned
How often energization slipped, and what the typical cause was
The gap between "installation complete" on the GC schedule and "ready for commissioning" in the field, which is rarely zero
Where issues cluster by phase, and how many get reopened, which is where retest time hides
How many attempts integrated testing took before it passed

The scheduler can still keep the duration. But now they're keeping it on purpose, with the risk visible and documented, instead of by default. That's a completely different position for the owner to be in when the slip happens.

The uncomfortable part

Most Cx firms have this data. Very few can use it, and fewer still at the right time.

It lives in one CxAlloy instance per project, in spreadsheets that were built for that job, in closeout PDFs, and in the memories of whoever ran the site. Naming conventions differ from project to project. One owner's Level 3 is another firm's Level 2. Issue priorities mean different things on different jobs. The experience is there. The evidence is trapped.

So when an owner is choosing a commissioning agent, "have you done this before" is a fair question but an incomplete one. The better question is whether the firm can show you, from its own records, what actually happened on projects like yours. That's a much higher bar, and the firms that can clear it are worth more, in the schedule review and everywhere else.

What this looks like in practice

For owners and project managers, the fix is a scope change and a timing change. Put schedule review in the CxA contract alongside design review, and get the CxA under contract before the baseline is due, not after. If that isn't possible, name the person on the owner's team who will review the baseline through a commissioning lens until it is. Then bring the CxA back at each major update, because the schedule that got approved is never the schedule that gets built.

For the CxA, the fix starts with asking for the schedule at all. A lot of firms don't until they're already living in it. And comments that arrive after the baseline is locked are worth a fraction of comments that arrive before.

For both sides, agree early on a short list of commissioning milestones that belong in the master schedule: energization by lineup, startup by system, functional testing windows, and integrated testing as the event it really is. Not every checklist, not every test — just the handful of points where the Cx sequence and the construction sequence have to meet.

Where the data goes from here

Once the milestones are real, you can measure them. Planned energization against actual. Planned functional testing duration against actual. How much float was left when integrated testing started. Across one project that's a lessons-learned slide. Across ten it's a pattern, and the pattern is what lets the next schedule review start from evidence instead of memory.

That's the problem CxLens is built around: taking the commissioning records that already exist across projects and making them consistent enough to learn from. But the tool is downstream of the habit. The habit is getting the person who understands the commissioning sequence in front of the schedule before it's locked, and giving them something better than a hunch to bring with them.